1. The data related to airport: Fix the pickup/dropoff spot at the airport and discover potential patterns/phenomena of filtered data.

1.1 Dispersion of starting point or destination. We again view the choropleth map, by deviding regions of NYC into geographical blocks and counting the number of trips arriving at / starting from each blocks with the airport as its starting point / destination, respectively. In other words, we fix one side of trip as the airport and are interested in the other side. First we view the choropleth map of trips starting from the airport.

In the plot, the redder the block is, the larger number it represents; blue color represents smaller number and grey color means no counts of trips in this region. A clearly regional dispersion can be discerned from the choropleth map: Manhattan has the largest numbers of trips (especially midtown), folllowd by north Brooklyn and west Queens; Bronx and Staten Island tends to have the fewest trips. This quite agrees with the Taxi customers density in NYC from the whole dataset. Note that airports themselves also have large number of counts, which can be reasonably accounted by the considerable requirment for people going from one airport to another in NYC to transfer flight.

We also view the choropleth map with the airport as its destination.

It gives us similar regional distributing patterns of counts as the previous plot, except for generally fewer trips in total than the previos plot. A plausibly reason can account for this phenomenon: People just arrive at the city tend to be more tired than those who are about to leave the city, thus they would tend to prefer more comfortable and convenient ways of transportation (like Taxi) than other transportation substitute (like subway or buses).

1.2 Trips distribution along Time of day. We plots higtograms of number of trips in every hour of a day. Trips are grouped by two dimensions: whether it arrives at or starts from the airport, and which airport of JKF / LGA.

Consequently we get four histograms. For each histogram, the veritcal axis denotes time in 24-hour-notation, and the horizontal axis denotes the number of trips. From the histogram, seveal interesing phenomena can be noticed:

  1. In plot (a), most people arrive at the airport evenly from 5am to 5pm, and very few people arrive at the airport form 9pm to 3am. In plot (c), the busy hours are from 8am to 11pm and the non-busy hours are from 1am to 6am. Considering the people usually arrive at the airport 1-2 hours before taking-off and leave the airport about 1 hour after landing, these two histograms properly reflects the density of flights in general.

  2. There are fewer late-night trips for LGA (plot a, c) than JFK (plot b, d). The number of trips starting from / arriving at JFK changes more smoothly than those of LGA. This may due to the fact that JFK mostly holds intenational flights with less stright late-night restrition than LGA, which mostly holds domestic flights.